Öznitelik Seçiminde Dilsel Kuvvetli Sinir Bulanık Sınıflayıcı Kullanımı

Yıl 2006, Cilt: 19 Sayı: 2, 109 - 130, 31.12.2006

Bayram Çetişli

Öz

Bu çalışmada, örüntü tanımanın basamaklarından biri olan öznitelik seçimi, bulanık kümelere uygulanan ve veriler üzerinden eğitilen dilsel kuvvetlerle yapılmaktadır. Bulanık kurallarda kullanılan “az”, “çok fazla” ve “çok az” gibi sıfatlar özniteliklerin sınıf için önemini ortaya koymaktadır. Buna göre ilk evrede her sınıf için en uygun olan ortak ve bireysel öznitelikler seçilmektedir. Seçilen öznitelikler ikinci evrede Dilsel Kuvvetli Sinir-Bulanık Sınıflayıcı (DKSBS) ile sınıflanarak başarımı ölçülmektedir. DKSBS ağ tabanlı bir sınıflayıcı olup, öznitelik-sınıf ilişkisini bulanık kurallarla çok iyi ortaya koyan bir yapıdır. Böylece ayrıştırmayı zorlaştıran gürültü, ölçüm hataları  içeren ya da ilgisiz olan öznitelikler elenerek, sınıflamada ayırt edici özelliği en iyi olan öznitelikler değerlendirilmeye alınmaktadır. 

Anahtar Kelimeler

Öznitelik seçimi, Dilsel kuvvetler, Sinir-bulanık sınıflayıcı. .

Usıng Neuro-Fuzzy Classıfıer Wıth Lıngustıc Hedges For Feature Selectıon

Öz

In this study, one of the important steps of pattern recognition is feature selection that can be made by linguistic hedges. The linguistic hedges are used in fuzzy rules and trained with supervised learning methods. Some adjectives such as “very”, “little”, “more or less”, and “rather” are used in fuzzy rules to reveal the importance of feature. According to these situations, the suitable discriminative features for every class should be selected in the first step. The selected features are classified using neuro-fuzzy classifier with linguistic hedges in the second step. Then the classification success is evaluated. In this way, noisy or irrelevant features are eliminated and, discriminative features in the classification are taken to evaluation. 

Anahtar Kelimeler

Feature selection, Linguistic hedges

Kaynakça

• [1] D. Huang and T. W. S. Chow, “Efficiently searching the important input variables using Bayesian discriminant” IEEE Trans. on Circuits and Systems-I: Regular Papers, Vol. 52, No. 4, pp. 785–793, 2005.
• [2] H.Liu, E.R. Dougherty, J.G. Dy, K. Torkkola, E. Tuv, H. Peng, C. Ding, F. Long, M. Berens, L. Parsons, Z. Zhao, L. Yu, G. Forman, “Evolving
• feature selection”, IEEE Intelligent Systems, Vol 20, Is. 6, pp. 64–76, 2005.
• [3] S. Abe, R. Thawonmas, and Y. Kobayashi, “Feature selection by analyzing class regions approximated by ellipsoids”, IEEE Trans. on Systems, Man, and Cybernetics-Part C: Applications and Reviews, Vol. 28, No. 2, pp. 282–287, 1998.
• [4] S. Sural, P.K.Das, “A Genetic Algorithm for Feature Selection in a Neuro-Fuzzy OCR System”, Document Analysis and Recognition, 2001. Proceedings. Sixth International Conference on 10–13 Sept. 2001, pp. 987 – 991.
• [5] J.M. Benitez, J.L. Castro, C.J. Mantas, and F. Rojas, “A Neuro-Fuzzy Approach for Feature Selection”, IFSA World Congress and 20th NAFIPS International Conference, 2001. Joint 9th, Vol. 2, 25–28 July 2001, pp. 1003 – 1008.
• [6] D. Chakraborty and N. R. Pal, “A Neuro-Fuzzy Scheme for Simultaneous Feature Selection and Fuzzy Rule-Based Classification”, IEEE Trans. on Neural Networks, Vol. 15, No. 1, pp. 110–123, 2004.
• [7] C. Emmanouilidis, A. Hunter, J. MacIntyre, C. Cox, “Multiple-criteria genetic algorithms for feature selection in neuro-fuzzy modelling”, Neural Networks, 1999. IJCNN '99. International Joint Conference on Vol. 6, 10–16 July 1999, pp. 4387–4392.
• [8] J.G. Marin-Blazquez, Q. Shen, “Linguistic hedges on trapezoidal fuzzy sets: a revisit”, Fuzzy Systems, 2001. The 10th IEEE International Conference on Vol. 1, 2–5 Dec. 2001, pp. 412 – 415.
• [9] J. Casillas, O. Cordón, M. J. del Jesus, and F. Herrera, “Genetic Tuning of Fuzzy Rule Deep Structures Preserving Interpretability and Its Interaction With Fuzzy Rule Set Reduction”, IEEE Trans. on Fuzzy Systems, Vol. 13, No. 1, pp. 13–29, 2005.
• [10] J. G. Marín-Blázquez, and Q. Shen, “From Approximative to Descriptive Fuzzy Classifiers”, IEEE Trans. on Fuzzy Systems, Vol.10, No. 4, pp. 484–497, 2002.
• [11] J.-S. R. Jang, C. T. Sun, E. Mizutani, Neuro-Fuzzy and Soft Computing, Prentice Hall, Upper Saddle River, 1997. [12] C.E. Shannon, “A symbolic analysis of relay and switching circuits”, Trans. American Institute of Electrical Engineers, Vol. 57, pp. 713–723, 1938.
• [13] L. Comtet, “Boolean Algebra Generated by a System of Subsets” in Advanced Combinatorics: The Art of Finite and Infinite Expansions, rev. enl. ed. Dordrecht, Netherlands: Reidel, pp. 185-189, 1974.
• [14] L. A. Zadeh, “The concept of a linguistic variable and its application to approximate reasoning” Parts 1,2 and 3, Information Sciences, 8:199– 249,8:301–357,9:43–80, 1975.
• [15] L. A. Zadeh, “Quantitative fuzzy semantics”, Information Sciences, 3:159-176, 1971.
• [16] C.-T. Sun and J.-S. R. Jang, “A neuro-fuzzy classifier and its applications”, Proceedings of IEEE International Conference on Fuzzy Systems, San Francisco, Vol. 1, 1993, pp. 94–98.
• [17] M. Møller, “A scaled conjugate gradient algorithm for fast supervised learning”, Neural Networks, Vol. 6, No. 4, pp.525–533, 1993. [18] UCI Machine Learning Group, www.ics.uci.edu/~mlearn.